[SCIP] Variable repeated

[Marc Pfetsch]

Hi Mahdi,

have you double checked that variable bounds are represented in your 
dual? If you have bounds on the variables you are pricing (i.e., 
potentially exponentially man) you should try to avoid them by using 
constraints that implicitly bound the variables. And finally, it makes 
sense to think about the signs of the coefficients in the dual (this has 
always been the problem for me when running into the same problem).

As a check you can compute the reduced costs for variables that already 
appear in your problem and compare them to the ones that SCIP gives you.

Best

Marc


On 13/03/2023 23:27, m.noorizadegan at gmail.com wrote:
> Hi Scip community,
> 
> I have a pure linear program, (no integer or binary variable).
> 
> The branch and price algorithm keeps identifying the same variables as 
> improving variables.
> 
> I thought there could be potentially two issues: 1) reduced cost is not 
> correctly computed and 2) new variables are not correctly assigned to 
> constraints.
> 
> I have double checked both many times. Seems these are not the issue.
> 
> I was wondering if someone could tell me if there are any other issues 
> that would lead to repeated variables?
> 
> We look forward to hearing back from you soon.
> 
> Best,
> 
> Mahdi
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip